基于K-means聚类的路面裂缝分割算法  

Pavement crack segmentation based on K-means clustering

作　　者：李鹏[1,2] 李强[1,2,3] 马味敏 蒋威 LI Peng;LI Qiang;MA Wei-min;JIANG Wei(Jiangsu Key Laboratory of Meteorological Observation and Information Processing,Nanjing University of Information Science and Technology,Nanjing 210044,China;School of Electronic and Information Engineering,Nanjing University of Information Science and Technology,Nanjing 210044,China;Hangzhou Meteorological Bureau,Hangzhou 310001,China)

机构地区：[1]南京信息工程大学江苏省气象探测与信息处理重点实验室,江苏南京210044 [2]南京信息工程大学电子与信息工程学院,江苏南京210044 [3]杭州市气象局,浙江杭州310001

出　　处：《计算机工程与设计》2020年第11期3143-3147,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(41075115);江苏省第11批六大高峰人才基金项目(2014-XXRJ-006);江苏省重点研发计划社会发展基金项目(BE201569)。

摘　　要：针对传统分割算法在非均匀背景中存在低抗噪性及高复杂性的问题,将聚类分析和区域生长算法相结合,提出基于K-means聚类的路面裂缝区域生长分割算法。依据图像灰度像素特征进行裂缝目标聚类,融合裂缝几何纹理特征将聚类中心值作为种子点区域生长,经过形态学滤波优化处理完成精确分割。仿真结果表明,该算法在视觉上裂缝分割的准确率提高,查准率曲线和受试者工作特征曲线的评价结果也表明,该算法与传统边缘检测算法相比,具有环境适应能力强、识别准确率高以及性能稳定等优势。Aiming at the problem of low noise resistance and high complexity of traditional segmentation algorithm in heteroge-neous background,combining clustering analysis with region growing algorithm,a segmentation algorithm of pavement crack region growth based on K-means clustering was proposed.The fracture targets were clustered according to the gray pixel features,the geometric texture feature regions of the fracture were fused for growth,and the accurate segmentation was completed through morphological filtering and optimization.Experimental results show that the accuracy of crack segmentation is improved in vision.The evaluation results of the precision-recall curve(PR)and the receiver operating characteristic curve(ROC)of the subjects show that the algorithm has the advantages of strong adaptability to the environment,high recognition accuracy and stable performance compared with the traditional edge detection algorithm.

关 键 词：裂缝检测 K-MEANS聚类 区域生长 查准率曲线 受试者工作特征曲线 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]